Image Reconstruction Scheme for Watersheds based Video Segmentation

In this paper, we proposed a new solution to the reconstruction of image in order to solve the oversegmentation problem of watershed transformation which is widely used as an image segmentation method. The process for reconstruction of image consists of consolidation of edges and removal of local minima. The former(consolidation of edges) enhances the edge between the object in image background and the other(removal of local minima) erodes partially an image in order to eliminate the small local minima. Comparing to the existing solution, the proposed method could efficiently solve the over-segmentation problem within a very short time. And, using the modified similarity measure, we could acquire the result which is independently merged even though the criterion area is different. So, our result can be applied real-time video based application such as video conference, surveillance and MPEG-4 based systems.